An algorithm for training artificial neural networks (ANNs) by adjusting their weights and biases to minimize a loss function, which measures the difference between the predicted and actual output values.

• Machine learning
• Artificial intelligence
• Deep learning
• Neural networks
• Training algorithms

Backpropagation is commonly used in image classification tasks, where an ANN takes an input image and outputs a class label. The algorithm propagates the error from the predicted class label backward through the network to adjust the weights of each neuron, improving the accuracy of the model over time.

• Managing and evolving the semantic data model